Google says reducing emissions is tricky because of AI’s growing demands

  • In its 2024 Environmental Report, Google notes that emissions stemming from energy consumption at the firm have increased 13 percent year-on-year.
  • This as the compute power required for AI grows in an increasingly competitive market.
  • As such, Google says reducing emissions while it pursues AI will be challenging.

Why let a little thing like your impact on the wider world get in the way of securing a piece of a market worth a potential $15.7 trillion in six years?

This appears to be the attitude of Google which noted that its emissions stemming from its energy consumption for the last year increased 13 percent compared to 2022, because of “increases in data centre energy consumption and supply chain emissions.” These data centres are largely where Google powers its generative artificial intelligence efforts.

The bad news is, that Google doesn’t see these emissions coming down anytime soon.

“As we further integrate AI into our products, reducing emissions may be challenging due to increasing energy demands from the greater intensity of AI compute, and the emissions associated with the expected increases in our technical infrastructure investment,” Google writes in its 2024 Environmental Report.

In fact, Google goes on to argue that the definition of “net zero” is really constantly in flux and that may mean something different in a few years. Right now the definition of net zero means balancing out emissions with removals over a specified period. We don’t see how that changes but we also don’t have the power to just change definitions companies like Google does.

Right now, emissions from the electricity purchased to power Google’s data centres and offices accounts for 24 percent of its total carbon footprint. Google does try to use carbon-free energy solutions but the fact of the matter is it’s simply not enough.

“However, compared to 2022, our Scope 2 (market-based) emissions—which originate primarily from our data center electricity consumption—increased by 37%, despite considerable efforts and progress on carbon free energy. This was due to data center electricity consumption outpacing our ability to bring more CFE projects online, specifically in the United States and Asia-Pacific region, CFE contracts terminating prior to those projects becoming operational, and the current mismatch between our approach to CFE and the GHG Protocol’s Scope 2 guidance. In fact, despite achieving a 100%
global renewable energy match, our Scope 2 emissions have increased,” Google reports.

We remind you, once again, that this is all in service of pursuing an AI play which is why we don’t foresee the situation getting any better any time soon. Well, not unless the bubble pops.

There are growing concerns that the computing requirements of AI are placing strain on already strained electricity grids. An analyst at Goldman Sachs told CBS News that electricity demand could increase by as much as 160 percent thanks in part to AI.

While renewable energy is a way to address this demand, it’s a complex process especially when you consider the scale of the demand, the solution gets a bit more murky.

For example, Teraco is building a 120MW solar facility in the Free State but it needed to negotiate with Eskom to wheel that power to its facilities through the utility’s infrastructure. While Teraco is a big company, one can only imagine the size of the solar farm The Goog would need to build to power Gemini solely on renewable energy.

The company tries to explain how AI will help address climate challenges but doesn’t say how that will be done. In the meantime, the environment is suffering and all in pursuit of what boils down to chatbots that can now search the web and infringe on copyright laws.

It all just seems like a bit of a waste if you ask us. AI has a massive question mark over it at the moment.

There are real concerns about the impact the technology will have on jobs, especially jobs done by people who really need the money. Add the growing emissions from data centres to that mix and one starts to question whether we should be backing this technology as much as we have been.

Maybe, this whole AI thing needs a rethink before it triggers another climate change tipping point.

[Image – catazul from Pixabay]


About Author


Related News